Protected Heritage Stock and Population Redistribution in Spain: Multivariate Evidence from Four 2021 Mobile-Phone Mobility Snapshots

This study examines tourism-relevant population redistribution across 3214 Spanish mobility areas on four dates in 2021 using aggregated mobile-phone indicators. Because travel purpose is unobserved, the dates are treated as mobility snapshots rather than as a normal-year tourism series. Two descriptive principal components explained 77.58% of the variance in five non-redundant indicators, and non-metric multidimensional scaling recovered similar geometry. A repeated-measures test showed a large joint date effect, F(15, 3194) = 1630.85, p < 0.001, partial η2 = 0.885. A comparison of solutions for k = 2–4, based on 100 sub-sampling replications, supported three overlapping exploratory profiles: resident-dispersing (n = 873), broadly connected (n = 992), and summer-receiving (n = 1344). The original expert/AI-assisted heritage score failed external convergent validation and was withdrawn. For 2047 unambiguously mapped areas, it was replaced with a reproducible registry-linked stock of protected immovable heritage assets. In this mapped sub-sample, protected-asset stock was associated with greater summer retention, receiving connectivity, and net population gain. Partial redundancy analysis yielded an incremental R2 of 5.01% after population and autonomous-community controls (restricted-permutation p = 0.0005); this value fell to 4.07% with province controls and 3.57% with additional nonlinear and province-specific population adjustment. Because the crosswalk covered 63.8% of areas and disproportionately represented smaller areas, these estimates describe a modest, sample-conditional association rather than a national or causal effect; unmeasured confounding and local spatial dependence remain unresolved. The profiles suggest distinct monitoring priorities but require validation with denser, multi-year data beyond pandemic-affected 2021.